A computational study of the general lot-sizing and scheduling model under demand uncertainty via robust and stochastic approaches. (February 2018)

• Record Type: Journal Article
• Authors: Alem, Douglas; Curcio, Eduardo; Amorim, Pedro; Almada-Lobo, Bernardo
• Is Part Of: Computers & operations research. Volume 90(2018)
• Page Start: 125
• Page End: 141
• Publication Date: 2018-02
• Subjects: Lot sizing and scheduling problems -- GLSP -- Robust optimization -- Stochastic programming -- Empirical study -- Monte carlo simulation; Operations research -- Periodicals; Electronic digital computers -- Periodicals; 004.05
• DOI: 10.1016/j.cor.2017.09.005
• Languages: English
• ISSNs: 0305-0548

Abstract

Highlights: We present the General Lot-Sizing and Scheduling Problem under uncertainty. For the first time, we study a budget-uncertainty set robust optimization model. A scenario-based two-stage stochastic programming model is also analyzed. The benefit of each approach is assessed via a Monte Carlo simulation procedure. We provide guidelines to select the most suitable model based on managers' preferences. Abstract: This paper presents an empirical assessment of the General Lot-Sizing and Scheduling Problem (GLSP) under demand uncertainty by means of a budget-uncertainty set robust optimization and a two-stage stochastic programming with recourse model. We have also developed a systematic procedure based on Monte Carlo simulation to compare both models in terms of protection against uncertainty and computational tractability. The extensive computational experiments cover different instances characteristics, a considerable number of combinations between budgets of uncertainty and variability levels for the robust optimization model, as well as an increasing number of scenarios and probability distribution functions for the stochastic programming model. Furthermore, we have devised some guidelines for decision-makers to evaluate a priori the most suitable uncertainty modeling approach according to their preferences.